Reforming of oxygenates for H2 production: Correlating reactivity of ethylene glycol and ethanol on Pt(111) and Ni/Pt(111) with surface d-band center

Article abstract of DOI:10.1021/jp0548927

The dehydrogenation and decarbonylation of ethylene glycol and ethanol were studied using temperature programmed desorption (TPD) on Pt(111) and Ni/Pt(111) bimetallic surfaces, as probe reactions for the reforming of oxygenates for the production of H₂ for fuel cells. Ethylene glycol reacted via dehydrogenation to form CO and H₂, corresponding to the desired reforming reaction, and via total decomposition to produce C_(ad), O_(ad), and H₂. Ethanol reacted by three reaction pathways, dehydrogenation, decarbonylation, and total decomposition, producing CO, H ₂, CH₄, C_(ad), and O_(ad). Surfaces prepared by deposition of a monolayer of Ni on Pt(111) at 300 K, designated Ni-Pt-Pt(111), displayed increased reforming activity compared to Pt(111), subsurface monolayer Pt-Ni-Pt(111), and thick Ni/Pt(111). Reforming activity was correlated with the d-band center of the surfaces and displayed a linear trend for both ethylene glycol and ethanol, with activity increasing as the surface d-band center moved closer to the Fermi level. This trend was opposite to that previously observed for hydrogenation reactions, where increased activity occurred on subsurface monolayers as the d-band center shifted away from the Fermi level. Extrapolation of the correlation between activity and the surface d-band center of bimetallic systems may provide useful predictions for the selection and rational design of bimetallic catalysts for the reforming of oxygenates.

Full text of DOI:10.1021/jp0548927

1686
J. Phys. Chem. B 2006, 110, 1686-1694
Reforming of Oxygenates for H2 Production: Correlating Reactivity of Ethylene Glycol and
Ethanol on Pt(111) and Ni/Pt(111) with Surface d-Band Center
Orest Skoplyak, Mark A. Barteau, and Jingguang G. Chen*
Center for Catalytic Science and Technology, Department of Chemical Engineering, UniVersity of Delaware,
Newark, Delaware 19716
ReceiVed: August 29, 2005; In Final Form: NoVember 29, 2005
The dehydrogenation and decarbonylation of ethylene glycol and ethanol were studied using temperature
programmed desorption (TPD) on Pt(111) and Ni/Pt(111) bimetallic surfaces, as probe reactions for the
reforming of oxygenates for the production of H2 for fuel cells. Ethylene glycol reacted via dehydrogenation
to form CO and H2, corresponding to the desired reforming reaction, and via total decomposition to produce
C
(ad), O(ad), and H2. Ethanol reacted by three reaction pathways, dehydrogenation, decarbonylation, and total
decomposition, producing CO, H2, CH4, C(ad), and O(ad). Surfaces prepared by deposition of a monolayer of
Ni on Pt(111) at 300 K, designated Ni-Pt-Pt(111), displayed increased reforming activity compared to
Pt(111), subsurface monolayer Pt-Ni-Pt(111), and thick Ni/Pt(111). Reforming activity was correlated with
the d-band center of the surfaces and displayed a linear trend for both ethylene glycol and ethanol, with
activity increasing as the surface d-band center moved closer to the Fermi level. This trend was opposite to
that previously observed for hydrogenation reactions, where increased activity occurred on subsurface
monolayers as the d-band center shifted away from the Fermi level. Extrapolation of the correlation between
activity and the surface d-band center of bimetallic systems may provide useful predictions for the selection
and rational design of bimetallic catalysts for the reforming of oxygenates.
I. Introduction
temperature and time. Deposition of monolayer coverage Ni
on Pt(111) at a substrate temperature of 600 K produces a
Production of hydrogen for use in fuel cells can be achieved
by selective reforming of oxygenates. The oxygenates may be
derived from renewable biomass and offer advantages such as
low toxicity, low reactivity, and compatibility with the current
infrastructure for transportation and storage. Recently, aqueous
phase reforming of oxygenates, such as ethylene glycol, has
been studied on various transition metals.1-11 The rate of
reforming decreased as follows: Pt, Ni > Ru > Rh, Pd > Ir,
while hydrogen selectivity showed the following trend: Pd >
Pt > Ni > Ru > Rh.3 Overall, Pt was identified as the most
promising catalyst. Nickel-based catalysts were found to be
active for reforming of ethylene glycol but also displayed
decreased hydrogen selectivity due to alkane production.3 Other
studies have investigated the replacement of precious metal
catalysts with less expensive substitutes, and a Sn/Ni bimetallic
catalyst has been reported with hydrogen selectivity comparable
to Pt/Al2O3.4,8-10 The reforming of other oxygenates, such as
ethanol, has been studied on bimetallic catalysts such as Rh-
Pt,12 Au-Rh,13 and Ni-Rh.14
Bimetallic catalysts offer one approach to altering the physical
and chemical properties of metals, in some cases producing
unique catalytic properties not seen for either of the parent
metals.15-19 Recent work has demonstrated that combined
experimental and theoretical studies can be used to understand
the reactivity of bimetallic alloys and to predict properties of
new combinations.20-32
The Ni/Pt bimetallic system displays novel catalytic properties
and has been studied extensively.22-27 The surface morphology
of a Ni/Pt(111) bimetallic surface depends on the deposition
surface with nickel atoms diffusing into the subsurface region,
leaving the top layer enriched with platinum, as confirmed by
STM, LEIS, XPS, and AES.23 The subsurface monolayer alloy,
designated as Pt-Ni-Pt(111), can be prepared by depositing
either Ni onto Pt(111)25 or Pt onto Ni(111)24 at 600 K at a
monolayer Ni or Pt coverage. Longer deposition times of Ni
onto Pt(111) at 600 K produce surfaces composed of several
nickel layers, which begin to approximate the structure and
reactivity of the Ni(111) surface. On the other hand, the
deposition of monolayer Ni at 300 K leads to the Ni atoms
residing on the surface, and this surface is designated as Ni-
Pt-Pt(111). Annealing the surface monolayer Ni-Pt-Pt(111)
prepared at 300 K to 600 K causes the Ni atoms to diffuse into
the subsurface and produces a Pt-Ni-Pt(111) subsurface
monolayer with similar structure and reactivity as those
deposited at 600 K.23
The Ni/Pt(111) bimetallic surfaces display chemical properties
different from those of either Pt(111) or Ni(111). For example,
on Pt-Ni-Pt(111), the desorption temperature of hydrogen
decreases significantly as compared to either Pt(111) or Ni-
(111).22 Since the dissociative adsorption of hydrogen is
nonactivated, this suggests a weaker metal-H bond on the
subsurface monolayer Pt-Ni-Pt(111). The availability of more
weakly bonded hydrogen increases the activity for low-
temperature hydrogenation of molecules such as cyclohex-
ene,22,24,27 1-hexene, and 1-butene.28 Other temperature pro-
grammed desorption (TPD) experiments have shown that the
decomposition of ethylene on Pt-Ni-Pt(111) is decreased
compared to Pt(111) or Ni(111).22,26 On the basis of HREEL
spectra, this difference has been attributed to the formation of
a weakly π-bonded intermediate on the bimetallic surface vs a
* To whom correspondence should be addressed. E-mail: chen@
che.udel.edu.
10.1021/jp0548927 CCC: $33.50 © 2006 American Chemical Society
Published on Web 01/10/2006

Reforming of Oxygenates for H2 Production
J. Phys. Chem. B, Vol. 110, No. 4, 2006 1687
more strongly di-σ-bonded intermediate on the parent metals.22,26
In contrast, surface monolayer Ni-Pt-Pt(111) displays very
different properties. The desorption temperature of H2 from Ni-
Pt-Pt(111) increases significantly compared to that from Pt-
(111) and is slightly higher than that from Ni(111).23 This
indicates a stronger metal-H bond on Ni-Pt-Pt(111) compared
to either of the parent metals.
Theoretical calculations confirm these experimental observa-
tions. The hydrogen binding energy on Pt-Ni-Pt(111) calcu-
lated by DFT is lower than that on Pt(111) or Ni(111).23 On
the other hand, on Ni-Pt-Pt(111), the hydrogen binding energy
calculated by DFT is higher than that on either of the parent
metals. The origin of this effect was attributed to the change in
the d-band of the bimetallic surfaces, as characterized by the
surface d-band center. For Pt-Ni-Pt(111), the d-band center
decreases compared to Pt(111) or Ni(111), shifting away from
the Fermi level, resulting in weaker interactions with adsorbates.
On the other hand, on Ni-Pt-Pt(111), the d-band center moves
closer to the Fermi level thereby increasing interactions with
adsorbates. A nearly linear correlation between the hydrogen
binding energy and the d-band center was established for a wide
range of bimetallic surfaces,29-31 consistent with the theory of
chemisorption by Hammer and Nørskov.32
In the current work, we investigate reactivity trends for the
reforming of oxygenates, such as ethylene glycol and ethanol,
on Pt(111) and Ni/Pt(111) bimetallic surfaces. The reactions of
ethylene glycol have been studied previously on Cu(110),33 Ni-
(100),34 Ag(110),35,36 Rh(111),37 and Mo(110),38 but to our
knowledge, surface science studies on Pt(111) or Ni/Pt(111)
have not been performed. Ethanol has been studied previously
on many surfaces, including Pt(111)39-41 and Ni(111),42,43 but
not on Ni/Pt(111) bimetallic surfaces. TPD results for ethylene
glycol and ethanol on Pt(111) and Ni/Pt(111) will be presented
first, followed by correlations of the activities with the d-band
center of the surfaces and predictions based on the surface
d-band center of other Pt-3d metal bimetallic surfaces.
directional tubing. In all experiments, the gas exposures were
made at a crystal temperature of 200 K through a 1/4 in. stainless
steel tube. For ethylene glycol experiments, dosing was per-
formed with the crystal several inches away facing the dosing
tube. Ethanol experiments were performed with the crystal
facing away from the dosing tube.
The Pt(111) surface was cleaned by repeated cycles of Ne+
sputtering at 600 K and annealing at 1100 K. Following the
last sputter cycle, 1 L of O2 at 890 K was used to remove carbon
left on the surface, followed by annealing at 1100 K for 5 min.
This cleaning procedure was repeated until negligible C or O
was detected by AES.
Ni/Pt(111) bimetallic surfaces were prepared by physical
vapor deposition (PVD) of Ni, as described earlier.22 The crystal
was held at either 600 or 300 K during Ni deposition, depending
on which surface was being prepared. The evaporative PVD
doser consisted of a tungsten filament with a high-purity Ni
wire (Alfa Aesar, 99.99+%) wrapped around it, mounted within
a tantalum enclosure. The Ni(849 eV)/Pt(241 eV) AES ratios
were used to determine Ni coverages, similar to procedures
described earlier where a Ni/Pt AES ratio of 1.0 was taken as
the monolayer coverage for Pt-Ni-Pt(111).22 For Ni-Pt-Pt-
(111), it is more difficult to determine the morphology of the
monolayer because Ni does not grow in a layer-by-layer mode
at 300 K. Therefore, for this work, Ni-Pt-Pt(111) was formed
by depositing Ni onto Pt(111) at 300 K for exactly the same
deposition time as for monolayer Pt-Ni-Pt(111). This proce-
dure leads to a difference in the AES Ni/Pt ratio between the
Pt-Ni-Pt(111) and Ni-Pt-Pt(111) surfaces. This is because,
for Pt-Ni-Pt(111), the subsurface Ni is somewhat screened
by the top layer of Pt. Thus, after monolayer deposition for
Ni-Pt-Pt(111), the AES ratios were approximately 1.5, as
compared to 1.0 for Pt-Ni-Pt(111). For the thick Ni/Pt(111),
deposition was carried out at 600 K until more than 6 ML of
Ni were deposited.
2.2. DFT Modeling. Theoretical calculations were performed
with the DACAPO package, which uses a plane wave basis set
with pseudopotentials to represent the core electrons. The PW91
functional was used for self-consistent electronic structure
calculations, with an electronic temperature of kBT ) 0.2 eV
for speeding up the convergence, extrapolated back to 0 K. The
surfaces were approximated by a 2 × 2 unit cell with three
layers of metal separated by five equivalent layers of vacuum,
repeated periodically. The top layer was relaxed to the lowest
energy configuration. A plane wave cutoff of 350 eV was used
for all calculations. The d-band density of states was determined
by projection of the plane waves onto spherical harmonic
orbitals. The d-band center was calculated as the first moment
of the projected d-band density of states on the surface atoms
relative to the Fermi level. The Pt-Ni-Pt(111) surface was
approximated by replacing the second layer Pt atoms with Ni,
while for Ni-Pt-Pt(111) the topmost layer was replaced with
Ni. The d-band center of the thick Ni/Pt(111) surface was
approximated as that for Ni(111). For other Pt-3d bimetallic
surfaces, the topmost layer was replaced with the 3d metal of
interest.
II. Experimental and Theoretical Methods
2.1. Techniques. The ultrahigh vacuum (UHV) chamber used
for the experiments was a two-level stainless steel chamber (base
pressure of 1 × 10-10 Torr), which was equipped with Auger
electron spectroscopy (AES) and TPD. In TPD experiments,
the samples were heated at a linear rate of 3 K/s. The opening
of the random flux shield of the quadrupole mass spectrometer
(UTI 100C) was placed at a distance of ∼5 mm from the sample
surface. The TPD experimental setup allowed up to 12 masses
to be monitored simultaneously. The Pt(111) single crystal
(Princeton Scientific, 99.99%) was 1 mm thick, with a 10 mm
diameter, and was oriented to within 0.5°. The crystal was spot
welded directly to two tantalum posts that served as electrical
connections for resistive heating, as well as thermal contacts
for cooling with liquid nitrogen. With this mounting scheme,
the temperature of the crystal could be varied between 100 and
1100 K. The sample temperature was measured by spot welding
a chromel-alumel thermocouple to the top edge of the Pt(111)
single crystal.
Ethylene glycol (Sigma-Aldrich, 99.8%) and ethanol (Sigma-
Aldrich, 99.5+%) were transferred into glass sample cylinders
and purified using repeated freeze-pump-thaw cycles. All
other gases were of research purity and were used without
further purification. The purity of all reagents was verified in
situ by mass spectrometry prior to use. Doses are reported in
langmuirs (1 L ) 1 × 10-6 Torr s) and are uncorrected for ion
gauge sensitivity or flux enhancement due to dosing with the
III. Results
3.1. TPD of Ethylene Glycol. Figure 1a displays TPD spectra
of ethylene glycol from Pt(111) following different initial
exposures. Ethylene glycol desorbed from Pt(111) in two peaks.
The higher temperature peak at 267 K saturated by 0.2 L and
was attributed to monolayer desorption. The onset of a lower
temperature peak at ∼224 K was observed at 0.2 L exposure
and was attributed to multilayer desorption.

1688 J. Phys. Chem. B, Vol. 110, No. 4, 2006
Skoplyak et al.
Figure 1. TPD spectra of (a) ethylene glycol, (b) H2, (c) CO, and (d) CH4 following different ethylene glycol exposures on Pt(111).
Ethylene glycol reacted on Pt(111) to form H2 and CO. Parts
b and c of Figure 1 display H2 and CO TPD spectra,
respectively, following different exposures of ethylene glycol
on Pt(111). H2 desorption occurred at 329 K following an
ethylene glycol exposure of 0.05 L. With increasing exposures,
the desorption temperature of H2 shifted to slightly lower
temperatures. The evolution of H2 was most likely desorption
limited, as the peak temperature was the same as that after
exposure of H2 from calibration experiments not shown here,
as well as literature data.22 CO desorption occurred at ap-
proximately 460 K, also in agreement with our own experiments
after exposure to CO and with literature data for a desorption-
limited process on Pt(111).44,45 The TPD spectra marked 0 L
show the contributions to H2 and CO spectra from background
adsorption. For H2, background adsorption was minimal, but
background adsorption of CO was significant. Both were
accounted for in quantifying H2 and CO yields, as explained
below. Figure 1d shows the 16 amu TPD spectra, monitored to
determine whether ethylene glycol reacted to form methane.
Several peaks were observed; however, none could be attributed
to methane. The lower temperature peaks at 228 and 267 K
occurred at the same temperature as ethylene glycol. The higher
temperature peaks at approximately 460 K occurred at the same
temperature as the peaks in the 28 amu spectra and were
attributed to the cracking of CO in the mass spectrometer.
Figure 2a contains TPD spectra of ethylene glycol from Ni/
Pt(111) surfaces following a saturation exposure of 0.2 L
ethylene glycol. Ethylene glycol desorbed in several peaks from
the subsurface monolayer Pt-Ni-Pt(111). The peak at 227 K
was consistent with multilayer desorption based on desorption
from Pt(111). Monolayer desorption of ethylene glycol occurred
from a peak centered at 257 K, with a high-temperature
shoulder. From the surface monolayer Ni-Pt-Pt(111) and the
thick Ni/Pt(111) surfaces, multilayer and monolayer desorption
were not well resolved, suggesting that desorption from the
monolayer decreased significantly. When the bimetallic surfaces
are compared to Pt(111), the desorption temperature of ethylene
glycol decreased slightly and the amount of chemisorbed
molecules desorbed intact decreased significantly on Ni-Pt-
Pt(111) and on the thick Ni/Pt(111) surfaces.
Parts b and c of Figure 2 show TPD spectra of H2 and CO,
respectively, following a saturation exposure of 0.2 L ethylene

Reforming of Oxygenates for H2 Production
J. Phys. Chem. B, Vol. 110, No. 4, 2006 1689
Figure 2. TPD spectra of (a) ethylene glycol, (b) H2, (c) CO, and (d) CH4 following 0.2 L ethylene glycol on Pt(111) and Ni/Pt(111) surfaces.
glycol from Ni/Pt(111) bimetallic surfaces. H2 desorbed from
Pt-Ni-Pt(111) as a weak, broad peak centered at approximately
300 K, a slightly lower temperature compared to Pt(111). CO
desorption occurred at 401 K, about 60 K lower than that from
Pt(111). The decrease in the desorption temperature of CO was
consistent with a desorption-limited process, as the heat of
adsorption of CO should decrease on Pt-Ni-Pt(111) due to
the decrease in the d-band center, as described in more detail
later. Both H2 and CO peak areas were reduced compared to
Pt(111). From Ni-Pt-Pt(111), H2 and CO desorbed from
multiple overlapping peaks. H2 desorbed at 360 and 385 K, with
a smaller third peak appearing as a low-temperature shoulder
at 327 K. CO desorption occurred at 429 and 456 K, with a
low-temperature shoulder at 401 K. Also, a fourth, much smaller
peak was observed at 512 K. The peak areas of H2 and CO
increased significantly on the Ni-Pt-Pt(111) surface as
compared to Pt(111) or Pt-Ni-Pt(111). From thick Ni/Pt(111),
H2 desorption occurred in a peak centered at 327 K, with a
high-temperature shoulder at 360 K. CO desorbed from two
overlapping peaks at 401 and 429 K, with a slight high-
temperature shoulder. Similar to Ni-Pt-Pt(111), the H2 and
CO peak areas from thick Ni/Pt(111) increased as compared to
Pt(111) or Pt-Ni-Pt(111). Figure 2d displays the 16 amu
spectra, which show that, similar to Pt(111), methane was not
formed on any of the bimetallic surfaces, as all the observed
peaks could be assigned to the cracking pattern of either ethylene
glycol or CO.
3.2. TPD of Ethanol. The reactions of ethanol exhibited both
similarities and differences relative to those of ethylene glycol.
Figure 3a displays TPD spectra of ethanol from Pt(111) after
different exposures of ethanol. Ethanol desorption occurred at
224 K. This peak saturated at an exposure of 0.3 L and was
attributed to monolayer desorption, consistent with previous
studies.39 A multilayer peak was not observed because the
dosing was performed while holding the crystal at 200 K to
prevent adsorption of water from the UHV background.
Although not shown here, separate experiments were performed
in which ethanol was dosed at 120 K, which showed that the
ethanol multilayer desorbed at 167 K, while water desorbed at
175 K, both consistent with literature results.40,41 Other than
the lack of water desorption, no changes in the TPD spectra
were observed for dosing at 200 K as compared to 120 K.

1690 J. Phys. Chem. B, Vol. 110, No. 4, 2006
Skoplyak et al.
Figure 3. TPD spectra of (a) ethanol, (b) H2, (c) CO, and (d) CH4 following different ethanol exposures on Pt(111).
Ethanol reacted on Pt(111) to form H2, CO, and CH4. Parts
b-d of Figure 3 show the H2, CO, and CH4 TPD spectra,
respectively, after different exposures of ethanol on Pt(111).
H2 desorption occurred at 355 K, independent of the initial
exposure of ethanol. CO desorbed from Pt(111) at 467 K, the
same temperature as CO desorption from ethylene glycol. CH4
desorption occurred at a peak temperature of 339 K. Methane
desorption was clearly reaction limited, as the desorption
temperature of molecularly adsorbed CH4 from Pt(111) has been
reported to be 67 K.46 The other peak at 466 K in the 16 amu
spectra was assigned to CO cracking by comparison with the
28 amu spectra. In general, the results presented here were in
good agreement with previous studies of ethanol on Pt(111).39-41
Figure 4a displays TPD spectra for ethanol from Ni/Pt(111)
bimetallic surfaces following a saturation exposure of 0.3 L
ethanol. From Pt-Ni-Pt(111) and the thick Ni/Pt(111) surfaces,
ethanol desorbed at approximately 230 K, just slightly higher
than from Pt(111). On the other hand, from Ni-Pt-Pt(111),
no ethanol desorption was observed. Parts b-d of Figure 4 show
the H2, CO, and CH4 TPD spectra, respectively. The reactivity
of ethanol on the Ni/Pt(111) bimetallic surfaces followed the
same trend as that of ethylene glycol. On Pt-Ni-Pt(111), the
reactivity of ethanol was lower compared to Pt(111). CH4 was
not detected, and H2 desorbed as a weak and broad peak centered
at approximately 300 K. CO desorbed at 404 K, similar to CO
desorption from the decomposition of ethylene glycol on this
surface and about 50 K lower than that from Pt(111). From
Ni-Pt-Pt(111), H2 desorbed from an intense peak at 367 K.
CO desorption occurred from three states, with the most intense
peak at 459 K, followed by two weaker peaks at 515 and 549
K. On this surface, a relatively intense CH4 peak was also
detected at 339 K, the same peak temperature as from Pt(111).
From thick Ni/Pt(111), H2 and CO desorbed at 363 and 425 K,
respectively. A small amount of CH4 was also detected at 339
K. The peak areas for H2 and CO increased significantly on
Ni-Pt-Pt(111) and thick Ni/Pt(111) surfaces as compared to
Pt(111) or Pt-Ni-Pt(111), similar to the behavior of ethylene
glycol on these surfaces.
IV. Discussion
4.1. Quantification of TPD Peak Areas. The saturation
coverage of H2 on Pt(111) is reported in the literature to be
∼0.45 ML (or 0.9 ML H(a))47 based on adsorption of D2 using

Reforming of Oxygenates for H2 Production
J. Phys. Chem. B, Vol. 110, No. 4, 2006 1691
Figure 4. TPD spectra of (a) ethanol, (b) H2, (c) CO, and (d) CH4 following 0.3 L ethanol on Pt(111) and Ni/Pt(111) surfaces.
nuclear microanalysis. The saturation coverage for CO on Pt-
(111) is reported to be 0.68 ML.48 In the current work, the
absolute coverages of the H2 and CO products were obtained
by comparing their TPD peak areas with those following the
saturation exposures of H2 or CO on Pt(111). The CH4 yields
were determined using an experimentally determined sensitivity
factor relative to CO by back filling the UHV system with equal
concentrations of CH4 and CO based on the ionization sensitivity
factors of the two gases. Product yields were then calculated
from the TPD results with appropriate subtraction of the
background contributions.
spectra taken after TPD confirmed the presence of C(ad)
following ethylene glycol exposures. The reforming activity,
a, and decomposition activity, b, are calculated as follows using
the H2 and CO yields from the TPD peak areas. CO is only
produced by reforming, and the value of a can be calculated
from the CO yield. H2 is produced in both reactions, thus the
value of b can then be determined using the H2 yield and the
value of a calculated from the CO yield.
1
2
CO yield ) 2a w a ) CO yield
The reaction pathways of ethylene glycol on Pt(111), Pt-
Ni-Pt(111), Ni-Pt-Pt(111), and thick Ni/Pt(111) can be
summarized as follows
1
3
H2 yield ) 3a + 3b w b ) H2 yield - a
Table 1 summarizes the calculated activities and selectivities
of ethylene glycol on Pt(111), Pt-Ni-Pt(111), Ni-Pt-Pt(111),
and thick Ni/Pt(111). The Ni-Pt-Pt(111) surface displays the
highest reforming activity and total activity. The reforming
activity of the thick Ni/Pt(111) is also significantly higher
compared to Pt(111). The Pt-Ni-Pt(111) surface shows the
lowest reforming and total activities. It is also interesting to
aHOCH2CH2OH f 2aCO + 3aH2
(1)
(2)
bHOCH2CH2OH f 2bC(ad) + 2bO(ad) + 3bH2
Reaction 1 is the desired reforming reaction, producing CO and
H2. Reaction 2 represents total decomposition, producing H2,
as well as C(ad) and O(ad) which remain on the surface. AES

1692 J. Phys. Chem. B, Vol. 110, No. 4, 2006
Skoplyak et al.
TABLE 1: Reactivity of Ethylene Glycol on Pt(111) and Ni/Pt(111) Surfaces
activity (ML)
selectivity (%)
surface
Pt(111)
Pt-Ni-Pt(111)
Ni-Pt-Pt(111)
thick Ni/Pt(111)
reforming
decomposition
total
reforming
decomposition
0.030
0.025
0.144
0.122
0.004
0
0.023
0.016
0.034
0.025
0.167
0.138
88
100
86
12
0
14
12
88
TABLE 2: Reactivity of Ethanol on Pt(111) and Ni/Pt(111) Surfaces
activity (ML)
selectivity (%)
surface
Pt(111)
Pt-Ni-Pt(111)
Ni-Pt-Pt(111)
thick Ni/Pt(111)
reforming
decomposition
CH4
total
reforming
decomposition
CH4
0.025
0.015
0.074
0.070
0
0
0.002
0
0.010
0.001
0.027
0.015
0.087
0.071
93
100
85
0
0
3
0
7
0
12
1
0.003
0
99
note that ethylene glycol reacts primarily through dehydroge-
nation and C-C bond scission on all four surfaces. The
selectivity toward total decomposition representing C-O bond
scission is at most 14%. This is important for hydrogen
production by reforming of oxygenates because C-O bond
scission is undesirable as it can lead to other products and
deposit C(ad) onto the surface, deactivating the catalyst.
The reaction pathways of ethanol on Pt(111), Pt-Ni-Pt(111),
Ni-Pt-Pt(111), and thick Ni/Pt(111) can be summarized as
follows
Similar to ethylene glycol, ethanol reacts primarily through
dehydrogenation and the desired C-C bond scission. C-O bond
scission is minimal and probably can be attributed to defect
sites on the surfaces, consistent with previous studies which
report negligible C-O bond scission in reactions of ethanol on
both Pt(111)39 and Ni(111).43 Theoretical studies of surface
reactions of ethanol and its derivatives also suggest that C-C
bond scission should be favored, as the activation barrier for
C-O bond scission is higher than that for C-C bond scission
on Pt(111).49,50
It is interesting to compare the reactivity of ethanol on thick
Ni/Pt(111) with literature studies on Ni(111).43 On thick Ni/
Pt(111), ethanol reacts to form primarily CO and H2, with small
amounts of CH4. However, on Ni(111), acetaldehyde (CH3CHO)
is also produced, in addition to CH4, CO, and H2. The
temperatures for the products common for both surfaces are
similar. Also, both surfaces show negligible C-O bond scission
during TPD experiments. Overall, thick (>3 ML) Ni/Pt(111)
and Ni(111) display similar but not identical properties, as the
thick Ni/Pt(111) surface still retains some bimetallic character.
Other studies have shown that thick bimetallic surfaces often
display reactivity different from the parent metals. For example,
thick (>3 ML) Pd/Mo(110) and Pt/W(110) surfaces exhibit H2
desorption temperatures different from pure Pd(111) and Pt-
(111),31 respectively. Also, the reactivity of ethylene on thick
Pt/Ni(111) surfaces is different compared to Pt(111).26 Further
work is needed to understand these differences.
aCH3CH2OH f aCO + 3aH2 + aC(ad)
bCH3CH2OH f bCO + bCH4 + bH2
cCH3CH2OH f 2cC(ad) + cO(ad) + 3cH2
(3)
(4)
(5)
Reaction 3 is the most desirable pathway for reforming, as it
produces CO, H2, and C(ad). The last of these can be removed
from the surface as CO by steam gasification with H2O or by
the Boudouard reaction with CO2 (C(ad) + CO2 T 2 CO).
Reaction 4 is a decarbonylation reaction, producing CH4 along
with CO and H2, which is undesired because 2 H2 molecules
are “lost” as methane. Reaction 5 represents total decomposition,
producing H2, as well as C(ad) and O(ad) which remain on the
surface. AES spectra taken after TPD confirmed the presence
of C(ad) following TPD experiments. The reforming activity, a,
decarbonylation activity, b, and total decomposition activity, c,
are calculated as follows using the H2, CO and CH4 yields. CH4
is only produced through reaction 2, and thus b equals the CH4
yield. CO is produced by reforming and decarbonylation. The
value of a is calculated using the CO yield and the value of b.
H2 is produced in all three reactions, and thus the value of c is
determined using the H2 yield along with the values of a and b.
4.2. Correlation and Prediction of Activity using the
Surface d-Band Center. To establish trends for different
bimetallic surfaces, the reactivity of ethylene glycol and ethanol
was correlated with the electronic properties of the correspond-
ing surfaces as characterized by their d-band centers. The d-band
center provides a useful measure of the electronic properties of
transition metal surfaces, as the adsorption energies of small
molecules have been shown to follow a nearly linear trend as
a function of the d-band center.32 The reforming activity of
ethylene glycol and ethanol on Pt(111) and Ni/Pt(111) can be
correlated with the surface d-band center, as shown in Figure
5. The correlation displays a fairly linear relationship, with
activity increasing as the d-band center moves closer to the
Fermi level. The reforming activity of ethylene glycol is higher
than that of ethanol, in agreement with studies on Rh(111).37,51
The selectivity to reforming for ethylene glycol and ethanol is
shown in Figure 6 and does not display any discernible trend
as a function of the d-band center. The reforming selectivity is
fairly constant within experimental error for all surfaces except
Pt-Ni-Pt(111) where it is slightly higher.
CO yield ) a + b w a ) CO yield - b
CH4 yield ) b w b ) CH4 yield
1
3
1
3
H2 yield ) 3a + b + 3c w c ) H2 yield - a -
b
Table 2 summarizes the calculated activities and selectivities
of ethanol on Pt(111), Pt-Ni-Pt(111), Ni-Pt-Pt(111), and
thick Ni/Pt(111). The reactivity trend of ethanol is very similar
to that of ethylene glycol. The Ni-Pt-Pt(111) surface displays
the highest reforming and total activities. The reforming activity
of ethanol on thick Ni/Pt(111) is increased compared to Pt-
(111). On the other hand, Pt-Ni-Pt(111) shows decreased
activity as compared to Pt(111).

Reforming of Oxygenates for H2 Production
J. Phys. Chem. B, Vol. 110, No. 4, 2006 1693
Figure 7. Prediction of ethylene glycol reforming activity on different
monolayer 3d-Pt bimetallic surfaces based on extrapolation from Ni/
Pt(111) surfaces.
Figure 5. Correlation of ethylene glycol and ethanol reforming activity
with the d-band center on Pt(111) and Ni/Pt(111) surfaces.
H2 dissociation is likely to be activated.31,53 The volcanic
behavior of the hydrogenation activity is consistent with
Sabatier’s principle that states there must be a balance in the
binding energy of adsorbed species for catalytic reactions to
take place.
In the current work, we consider Pt-3d bimetallic surfaces
where the 3d metals occupy the top layer. As shown in Figure
7, DFT modeling results indicate that the presence of surface
monolayer 3d metals on Pt(111) shifts the surface d-band center
closer to the Fermi level, therefore increasing the interactions
with adsorbates. By the use of the trend between activity and
d-band center established experimentally in Figure 5, Figure 7
displays an extrapolation of ethylene glycol reforming activity
on different 3d-Pt-Pt(111) bimetallic surfaces. A similar
predictive trend can be established for ethanol. The trend predicts
increased activity on surfaces with d-band centers closer to the
Fermi level, opposite of the trend observed for hydrogenation
reactions. The shift in the d-band of 3d-Pt-Pt(111) surfaces
increases as the 3d metal moves to the left of the periodic table.
If the established linear trend continues, then alloying Ti with
Pt(111) is predicted to produce a surface nearly twice as active
as Ni-Pt-Pt(111), the most active surface alloy in the present
study, and seven times as active as Pt(111). However, the linear
trend between predicted reactivity and the d-band center might
not continue and a volcanic-type behavior might be observed,
as reported for the hydrogenation activity of cyclohexene.53
Further experimental studies on Fe/Pt(111) and Ti/Pt(111) are
under way to investigate the reactivity of ethylene glycol and
ethanol on other 3d-Pt-Pt(111) bimetallic surfaces to confirm
the prediction from the current study.
4.3. Stability of the Ni-Pt-Pt(111) Surface. Although the
Ni-Pt-Pt(111) surface shows much higher reforming activity
than Pt-Ni-Pt(111), experimental studies23 and DFT model-
ing23,29 indicate that monolayer 3d atoms prefer to reside in the
subsurface region under UHV or reducing conditions. However,
recent experimental and DFT studies54 show that the surface
monolayer is more thermodynamically stable than the subsurface
monolayer in the presence of oxygen. For example, in the
presence of 10-8 Torr of O2, subsurface Ni atoms start to diffuse
on top of Pt(111) at 450 K, while the diffusion of Co in Pt-
Co-Pt(111) occurs at temperatures as low as 350 K.54 Because
Figure 6. Reforming selectivity of ethylene glycol as a function of
the d-band center on Pt(111) and Ni/Pt(111) surfaces.
From the experiments carried out in this work, the Ni-Pt-
Pt(111) surface with the monolayer Ni residing in the top layer
shows the highest reforming activity compared to either Pt-
(111), thick Ni/Pt(111), or subsurface monolayer Pt-Ni-Pt-
(111) surfaces. Using the experimentally determined linear trend
of activity vs the surface d-band center, one can explore the
reactivity of other Pt-3d bimetallic surfaces toward reforming
of ethylene glycol and ethanol. It is known that for Pt-3d metal
bimetallic surfaces, substituting a 3d metal in the subsurface of
Pt(111) tends to decrease the d-band center, resulting in weaker
interaction with adsorbed species such as hydrogen and oxy-
gen.29 Furthermore, the d-band center of the Pt-3d-Pt(111)
surfaces, as compared to Pt(111), shifts further away from the
Fermi level as the 3d metal moves to the left side of the periodic
table. For reactions such as hydrogenation, this is desirable as
the weaker binding of hydrogen is expected to increase the
hydrogenation activity.52 However, the increase in hydrogenation
activity does not continue for surfaces that exhibit very weak
binding of hydrogen, such as Pd/Mo(110) and Pt/W(110), where

1694 J. Phys. Chem. B, Vol. 110, No. 4, 2006
Skoplyak et al.
(9) Shabaker, J. W.; Dumesic, J. A. Ind. Eng. Chem. Res. 2004, 43,
3105.
(10) Shabaker, J. W.; Simonetti, D. A.; Cortright, R. D.; Dumesic, J.
A. J. Catal. 2005, 231, 67.
(11) Davda, R. R.; Shabaker, J. W.; Huber, G. W.; Cortright, R. D.;
Dumesic, J. A. Appl. Catal., B 2005, 56, 171.
oxygen-containing molecules, such as water, are present in the
reforming reaction environment and the reaction is carried out
at approximately 500 K,1-11 it is possible that the more active
3d surface monolayer might be present in the 3d-Pt bimetallic
catalysts under reforming conditions. More direct measurements,
such as careful EXAFS (extended X-ray adsorption fine
structure) studies under in situ reaction conditions, would reveal
whether the diffusion of subsurface 3d metals occurs during
reforming.
(12) Sheng, P. Y.; Yee, A.; Bowmaker, G. A.; Idriss, H. J. Catal. 2002,
208, 393.
(13) Sheng, P. Y.; Idriss, H. J. Vac. Sci. Technol., A 2004, 22, 1652.
(14) Kugai, J.; Velu, S.; Song, C. Catal. Lett. 2005, 101, 255.
(15) Goodman, D. W. Ultramicroscopy 1990, 34, 1.
(16) Berlowitz, P. J.; Houston, J. E.; White, J. M.; Goodman, D. W.
Surf. Sci. 1988, 205, 1.
(17) Berlowitz, P. J.; Goodman, D. W. Surf. Sci. 1987, 187, 463.
(18) Rodriguez, J. A. Surf. Sci. Rep. 1996, 24, 223.
(19) Sinfelt, J. H. Bimetallic Catalysts: discoVeries, concepts, and
applications; Wiley: New York, 1983.
(20) Greeley, J.; Mavrikakis, M. Nat. Mater. 2004, 3, 810.
(21) Hammer, B.; Nørskov, J. K. Nature 1995, 376, 238.
(22) Fru¨hberger, B.; Eng, J., Jr.; Chen, J. G. Catal. Lett. 1997, 45, 85.
(23) Kitchin, J. R.; Khan, N. A.; Barteau, M. A.; Chen, J. G.;
Yahskinskiy, B.; Madey, T. E. Surf. Sci. 2003, 544, 298.
(24) Khan, N. A.; Zellner, M. B.; Chen, J. G. Surf. Sci. 2004, 556, 87.
(25) Khan, N. A.; Hwu, H. H.; Chen, J. G. J. Catal. 2002, 205, 259.
(26) Khan, N. A.; Zellner, M. B.; Murillo, L. E.; Chen, J. G. Catal.
Lett. 2004, 95, 1.
(27) Hwu, H. H.; Eng, J., Jr.; Chen, J. G. J. Am. Chem. Soc. 2002, 124,
702.
(28) Murillo, L. E.; Khan, N. A.; Chen, J. G. Surf. Sci. 2005, 594, 27.
(29) Kitchin, J. R.; Nørskov, J. K.; Barteau, M. A.; Chen, J. G. J. Chem.
Phys. 2004, 120, 10240.
(30) Kitchin, J. R.; Nørskov, J. K.; Barteau, M. A.; Chen, J. G. Phys.
ReV. Lett. 2004, 93, 156801.
(31) Zellner, M. B.; Goda, A. M.; Skoplyak, O.; Barteau, M. A.; Chen,
J. G. Surf. Sci. 2005, 583, 281.
V. Conclusions
The reactions of ethylene glycol and ethanol have been
studied on Pt(111) and Ni/Pt(111) bimetallic surfaces, as probes
of the reforming of oxygenates for the production of H2 for
fuel cells. The following conclusions can be drawn:
(1) The surface monolayer Ni-Pt-Pt(111) displays the
highest reforming activity, compared to Pt(111), subsurface
monolayer Pt-Ni-Pt(111), and thick Ni/Pt(111) surfaces.
Reactions involving C-O bond scission are minimal on these
surfaces.
(2) A linear correlation between reforming activities of both
ethylene glycol and ethanol and the d-band center of these
surfaces has been established.
(3) DFT modeling has been performed to predict the
reforming activity of 3d-Pt-Pt(111) bimetallic surfaces with
other 3d metals as the surface monolayer. The modeling results
predict that the activity should increase as the d-band center of
the bimetallic surface moves closer to the Fermi level, which
can be achieved by choosing 3d metals from the left side of
the periodic table. The activity trend for the reforming of
oxygenates is the opposite of that observed for hydrogenation
reactions on Pt-3d bimetallic surfaces.
(32) Hammer, B.; Nørskov, J. K. AdV. Catal. 2000, 45, 71.
(33) Bowker, M.; Madix, R. J. Surf. Sci. 1982, 116, 549.
(34) Madix, R. J.; Yamada, T.; Johnson, S. W. Appl. Surf. Sci. 1984,
19, 43.
(35) Capote, A. J.; Madix, R. J. Surf. Sci. 1989, 214, 276.
(36) Capote, A. J.; Madix, R. J. J. Am. Chem. Soc. 1989, 111, 3570.
(37) Brown, N. F.; Barteau, M. A. J. Phys. Chem. 1994, 98, 12737.
(38) Queeney, K. T.; Arumainayagam, C. R.; Weldon, M. K.; Friend,
C. M.; Blumberg, M. Q. J. Am. Chem. Soc. 1996, 118, 3896.
(39) Lee, A. F.; Gawthrope, D. E.; Hart, N. J.; Wilson, K. Surf. Sci.
2004, 548, 200.
Acknowledgment. We acknowledge financial support from
the U.S. Department of Energy, Office of Science, Division of
Chemical Sciences (Grant FG02-03ER15468).
(40) Sexton, B. A.; Rendulic, K. D.; Hughes, A. E. Surf. Sci. 1982, 121,
181.
(41) Rendulic, K. D.; Sexton, B. A. J. Catal. 1982, 78, 126.
(42) Xu, J.; Zhang, X.; Zenobi, R.; Yoshinobu, J.; Xu, Z.; Yates, J. T.
Surf. Sci. 1991, 256, 288.
(43) Gates, S. M.; Russell, J. N., Jr.; Yates, J. T., Jr. Surf. Sci. 1986,
171, 111.
Note Added after ASAP Publication. The title has been
modified. This paper was published ASAP on 1/10/06. The
corrected version was reposted on 1/18/06.
(44) Kostov, K. L.; Jakob, P.; Menzel, D. Surf. Sci. 1997, 377-379,
802.
References and Notes
(45) Thomas, V. D.; Schwank, J. W.; Gland, J. L. Surf. Sci. 2000, 464,
153.
(46) Sloan, D. J. A.; White, J. M. Surf. Sci. 1996, 365, 212.
(47) Norton, P. R.; Davies, J. A.; Jackman, T. E. Surf. Sci. 1982, 121,
103.
(48) Ertl, G.; Neumann, M.; Streit, K. M. Surf. Sci. 1977, 64, 393.
(49) Gursahani, K. I.; Alcala´, R.; Cortright, R. D.; Dumesic, J. A. Appl.
Catal., A 2001, 222, 369.
(1) Shabaker, J. W.; Davda, R. R.; Huber, G. W.; Cortright, R. D.;
Dumesic, J. A. J. Catal. 2003, 215, 344.
(2) Shabaker, J. W.; Huber, G. W.; Davda, R. R.; Cortright, R. D.;
Dumesic, J. A. Catal. Lett. 2003, 88, 1.
(3) Davda, R. R.; Shabaker, J. W.; Huber, G. W.; Cortright, R. D.;
Dumesic, J. A. Appl. Catal., B 2003, 43, 13.
(4) Shabaker, J. W.; Huber, G. W.; Dumesic, J. A. J. Catal. 2004,
222, 180.
(5) Cortright, R. D.; Davda, R. R.; Dumesic, J. A. Nature 2002, 418,
964.
(6) Davda, R. R.; Dumesic, J. A. Chem. Commun. 2004, 1, 36.
(7) Davda, R. R.; Dumesic, J. A. Angew. Chem., Int. Ed. 2003, 42,
4068.
(50) Alcala´, R.; Mavrikakis, M.; Dumesic, J. A. J. Catal. 2003, 218,
178.
(51) Houtman, C. J.; Barteau, M. A. J. Catal. 1991, 130, 528.
(52) Khan, N. A.; Murillo, L. E.; Chen, J. G. J. Phys. Chem. B 2004,
108, 15748.
(8) Huber, G. W.; Shabaker, J. W.; Dumesic, J. A. Science 2003, 300,
2075.
(53) Zellner, M. B.; Chen, J. G. J. Catal. 2005, 235, 393.
(54) Hwu, H. H.; Menning, C. A.; Chen, J. G. Manuscript in preparation.